Related Experiment Video
Updated: Jan 16, 2026
05:37
Author Spotlight: A Non-Invasive Tool to Assess and Differentiate Fat Patterns in Liver Using 3D Dixon MRI
Published on: October 20, 2023
2.1K
Towards population scale testis volume segmentation in DIXON MRI
Jan Ernsting1, Philipp Nikolas Beeken2, Lynn Ogoniak2
1Institute for Geoinformatics, University of Münster, Münster, Germany; Faculty of Mathematics and Computer Science, University of Münster, Münster, Germany; University of Münster, Institute for Translational Psychiatry, Münster, Germany.
Computers in Biology and Medicine
|September 30, 2025
Summary
Machine learning models can accurately measure testis volume from MRI scans, aiding male fertility research. This study provides tools for large-scale, reproducible analysis of testicular size in populations.
Area of Science:
- Radiology
- Medical Imaging
- Machine Learning
Background:
- Testis size is a key indicator of male fertility, typically assessed clinically through palpation or imaging.
- Population-level analysis of testicular volume using imaging is limited due to data constraints.
- Previous research showed machine learning's potential for testis volume segmentation.
Purpose of the Study:
- To evaluate machine learning segmentation methods for testicular volume using UKBiobank Magnetic Resonance Imaging (MRI) data.
- To establish a reproducible framework for large-scale testis MRI segmentation.
- To provide a trained model, baseline methods, and annotated data for research accessibility.
Main Methods:
- Utilized Magnetic Resonance Imaging (MRI) data from the UKBiobank.
- Evaluated various machine learning models for automated testis volume segmentation.
- Compared model performance against human interrater reliability.
Main Results:
- The best performing model achieved a median Dice score of 0.89.
- This performance surpasses the median Dice score of 0.85 observed for human interrater reliability.
- Enabled the first large-scale annotation of testicular volume using automated methods.
Conclusions:
- Machine learning offers a robust and accurate method for testicular volume segmentation from MRI.
- The developed model and dataset facilitate unprecedented population-scale analysis of testicular size.
- Enhances reproducibility and accessibility in male reproductive health research.
